Glossary

This glossary defines the core concepts used throughout the Cerebel documentation.

For architecture-level context and relationships between these terms, see Why Cerebel.

Cerebel

The name of the overall project.

Cerebel is a working system built around a Cerebrum substrate together with the supporting code required to launch it, run it, and use it in a real system.

At a high level, a Cerebel can be described as a domain specialist.

Substrate

A substrate is a bootable or launchable system unit in this architecture, with its own declaration, acquisition source, launch contract, and lifecycle policy.

A substrate may be a whole system or a subsystem; examples include Cerebrum, cortices, Thalamus, and future additions.

In the semantic suffixing naming model, Substrate is also reserved as the suffix for fully assembled concrete runtime classes.

Lamina

A lamina is an inherited structural layer in a class stack.

A lamina is intentionally incomplete by itself and contributes one capability layer toward a final assembled substrate.

In naming, lamina classes use the Lamina suffix.

Cerebrum

The Cerebrum contains the core structures that make the system teachable and capable of neuroplasticity.

It is the central intelligence substrate inside a Cerebel.

It also serves as the top-level API surface for the domain expert system.

Cerebellum

The Cerebellum is the supervisory partner to the Cerebrum.

It is responsible for human-in-the-loop correction, monitoring, and feedback.

Thalamus

The Thalamus is a coordinating substrate that routes messages between Cortices.

It acts as the routing and exchange layer that allows specialized parts of the system to work together coherently.

Cortex

A Cortex is a specialized assembled substrate for a domain responsibility.

The reusable inheritance layers that contribute to a Cortex are expressed as laminae such as BaseLamina, CortexLamina, SensoryMotorCortexLamina, SensoryCortexLamina, and MotorCortexLamina.

Prefrontal Cortex

The Prefrontal Cortex is one of the currently required Cortices in a Cerebrum.

As in biological brains, it is responsible for much of the reasoning, orchestration, and processing of data coming in from different inputs and other Cortices.

SensoryMotorCortexLamina

The SensoryMotorCortexLamina is the shared communication-oriented lamina for a Cerebrum.

It holds the reusable behavior that lets a Cerebrum participate in input/output exchange, routing, and interface adaptation without being the final assembled runtime on its own.

RAS

The Reticular Activating System, or RAS, is the broadcast and boot orchestration layer for substrates.

It broadcasts substrate needs to registered synthesizers, collects the returned modulators, and configures the working system before runtime interaction begins.

See also: Cerebrum System Architecture

Signal

A Signal is a directed message that carries an intent or action toward a specific recipient.

In the NeuralRelay vocabulary, a Signal is the brain-inspired replacement for a software-style request.

Signal maps to a targeted neural transmission sent through a terminal, distributed by a relay, or passed substrate-to-substrate on an axis.

Reply

A Reply is the return message that acknowledges a Signal or completes its requested work.

In the NeuralRelay vocabulary, a Reply is the brain-inspired replacement for a software-style response.

Reply maps to a feedback-like return signal that confirms or resolves the original transmission.

Broadcast

A Broadcast is a diffuse message that is delivered to many recipients rather than one specific target.

In the NeuralRelay vocabulary, a Broadcast is the brain-inspired replacement for a software-style event.

Broadcast maps to a neuromodulatory release pattern that influences multiple regions at once.

Application Layer

The Application Layer is the outside-world boundary that sits in front of a Cerebel runtime.

It validates and normalizes requests from adapters, applies boundary policy, and delegates to RAS and the runtime through a stable interaction contract.

Multiple application-layer implementations may exist for the same Cerebel, such as a CLI application, service application, or robotics gateway.

Adapter

An Adapter translates between a tool-native or transport-native protocol and the normalized application boundary contract.

Adapters handle transport and device concerns, but they do not perform domain reasoning or direct substrate orchestration.

AssociationCortex

The AssociationCortex fuses signals from multiple modalities into a unified working representation.

It provides a dedicated place for multi-modal integration before executive decision-making.

See also: Cerebrum System Architecture

SensoryCortexLamina

A SensoryCortexLamina encodes raw external input into the system’s internal working representation.

Different implementations can exist for text, speech, vision, or future modalities, and each contributes one layer of the final assembled sensory substrate.

See also: Cerebrum System Architecture

MotorCortexLamina

A MotorCortexLamina decodes internal decisions into raw output suitable for an external consumer.

Different implementations can exist for text, speech, vision, or future modalities, and each contributes one layer of the final assembled motor substrate.

See also: Cerebrum System Architecture

NeuromodulatorySynthesizer

The NeuromodulatorySynthesizer responds to RAS broadcast requests by producing modulators that shift behavior across the system.

It is the factory surface for cross-cutting state signals rather than a direct decision-making region.

See also: Cerebrum System Architecture

Domain Bundle

A Domain Bundle is the folder-scoped markdown startup configuration unit for one substrate.

In the current layout it includes one domain root file and one or more cortex files:

config/domains/<domain_id>/cerebrum.md
config/domains/<domain_id>/cortices/*.md

Modulator

A Modulator is an injected, cross-cutting signal or dependency that changes how a substrate behaves without changing the substrate’s core identity.

In the brain analogy, this parallels a neuromodulator: a broadcast influence that shifts gain, context, or behavior across multiple regions.

In software terms, a Modulator is requested by RAS, produced by one or more synthesizers, injected into a ModulatorCocktail, and then read by one or more runtime layers.

Design constraints:

A Modulator is additive and behavioral, not identity-defining.
A Modulator may be static (set at launch time) or dynamic (updated at runtime).
A Modulator should be explicit in layer contracts and tests.

Semantic Suffixing

Semantic suffixing is a naming convention where class suffixes communicate architectural role.

In this model, names are intentionally explicit so the role is readable directly from the symbol.

Common role suffixes:

Lamina for inherited structural layers.
Substrate for assembled concrete runtime classes.
synthesizer for modulator factories.
Signal, Reply, and Broadcast for orchestration messages.
Modulator for runtime influences.
Cocktail for grouped modulator state.